Qubic Light's LED Projector Looks Good - Page 2

Can someone provide some resources for single-panel LCoS technologies? I am more than a little curious on this aspect of LCoS.

I have a few questions.

Let us assume r,g.b LEDs since this is what we are talking about.

If a singel chip lcos panel is used it will have to be a quick one to get rid of rainbows. Let us assume a 600Hz cycling frequency.

sxrd at 5ms not fast enough , medium high cr
Philips lcos at 1ms is fast enough, low cr
but the LEDs can be turned off so we can reporduce black

A theoretical projector
600Hz rgb cycling
r,g,b LEDs and light output at 200 ANSI lumen
600:1 cr lcos panel analog modulation, but wait a minute can we adjust the intensity of the r.g. b LEDs?
1280x720

Philips already has a one-panel LCOS RPTV. No reason it couldn't be adapted.

Intel's LCOS designed are apparently focused on single-chip applications, too.

Yeah, that's what we've said and heard about 1080P technologies being deployed in RPTV's, too. Unfortunately I am one of those anomalies with the 'seeing is believing' philosophies.

The one panel LCOS's on the market right now have horrible rainbows. The MicroVue is actually a faster LCoS chip that behaves much like a DMD in a pulse-width modulation manner.

It isn't quite as fast as the DLP from my understanding but it would be fast enough.

The light spectrum WILL be sequential RGB LED's. White LEDs will not be used. The only thing that makes it viable is that you can operate LEDs in burst mode at 3 times their normal output at a dutycycle of 33%.

But even with that clever form of illumination, it appears that the LED still aren't bright enough to deliver a bright image. My guess is we are talking about 50 lumens for his prototype and the production are going to have similar issues whenever they are released.

However, I've said it a million times that LED light once it is efficient enough is going to work really really well.

-Mr. Wigggles

MrWigggles
How do you see the future LEDs versus solid state diode pumped lasers?

I have asked before and I do so again, does anybody know the lumes per watt ratio for
ssdp lasers
LEDs
UHP lamps
Xenon lamps

50 lumens? That seems to suggest that a 10 to 20-fold improvement in LED light output would be necessary for front projection. Is this realistically achievable?

Dave

From the prototype presentation:


You would need 4 times the light energy to produce the same image on a 100" diagonal screen. Given that the screen material wasn't given (some gain I'm sure) and the remarks that it wasn't bright, and didn't have colors quite rights leads me to believe D65 (i.e. ~6500K) lumens were in the neighborhood of 50.

I think that production models will be 200 lumens which will work for HT if you design your theater correctly - modest size screen and HIGH gain. (Silverstar and Hi-power come to mind)

People need to remember that in a normal DLP, a 250 Watt UHP lamp will put 5000+ lumens through the input aperture and yet you'll be lucky for 1000 to come out. Now 2/3 rds of that light is lost by the colorwheel which for the most part won't happen with a LED based light source (however light will have to be combined through an x-block which is not lossless.

A well designed LED projector with today's LED's might get close 200 D65 lumens but it will 3+ years before by my fearless estimate that we will see something in the 800+ lumen range.

-Mr. Wigggles

If they can manage to do 300 lm for $1000, I'd be willing to buy 3 PJs and stack them.

I have 2 LED light bulbs. 20 white LEDs per bulb.
They rate these the same brightness as a 100 watt incandescent bulb but they only drain 4 watts.

They are not very bright at all as far as giving off ambient light and lighting up a room but if you look directly at these your blinded for some time. Way brighter then looking directly at a 100 watt bulb. I they can capture and focus the hot spot of each LED I can see where they would be bright enough.

The problem with LED's isn't that they aren't efficient. The problem is that they are bulky and that makes focusing the light much more difficult (unless you use a huge imager)

-Mr. Wigggles

Hey all,

I know there have been a few LED projector threads, but this seems to be the most recent.

Just stumbled onto NEC's patent, obtained in Nov of 2003, for a LED-based DMD projector: Patent 6644814

It uses "fly eye" lenses and a dichroic prism to control and combine the arrays of R,G, and B light sources.

[ Edit: The link may not always work, so here are two other ways of getting there: Another Search resolve link or Enter 6644814 here ]

DocDvd has tested a LED RGB panel projector. He said its not quit ready for prime time ( not very bright) but the contrast ratio was excellent because like a CRT the LEDs are on or off.

I have many LED flashlights. Currently my brightest one has 3 5 watt Luxeon Star LED's fitting in a mag light head. It runs on 6 D cell batteries for 10 hours or more. The flashlight puts out in excess of 300 lumens in a blinding focused beam. If I shine it at my screen with my projector on, it almost completely overwhelms the image.

It is not a question of if LED's can be used effectively in a projector, but when it will happen. With the advancements in the technology, I believe that this will be the future.

Alan G.
I guess you mean that LEDs can be anything from off to full current on. In that sense they should provide for adjusting the light output. The important point in your mind must be that they can be totally off or?

Sunnybrook works on high dynamic range (HDR) displays. They mention leds at 250000 cd per square meter!! They work on displays that can show daylight close to what it is 10000 cd per square meter. They work on a display with led backlighting and an lcd in front. The leds adjust light level on lower spatial frequency and the lcd displays the high spatial frequency information.

That NEC patent is fascinating.

What is also fascinating is that Luxeon published a white paper describing how to use LEDs in a projector about the same time as the NEC patent was filed. (maybe some prior art issues?)

Anyway, the NEC device still uses A LOT O' glass to obtain its goal which means it ain't going to be cheap to impliment. I count 19+ (probably ~25)lens in the design.

But the benefits of LED's if you can get the brightness high enough are staggering. I walk into Best Buy on occasion and the Samsung DLP's are always some of the dimmest displays shown. They don't start out that way but they always end up that way (yes, the bulbs are on a lot but so are the CRT projectors that are right next to them)

No moving parts and 100,000 or so runtime life? I would never turn my projector off. It would also probably only be 120 watts of total dissipation.

The possibilities are incredible, they just need to get the lumens up.

-Mr. Wigggles

Ps. edit: here is a link to the 10/27/2002 SID presentation about Luxeons in projectors. http://www.lumileds.com/pdfs/techpap...BA-Paolini.PDF. I can not find the whitepaper about this though. I don't know too much about the patent process to know if this constitutes prior art or how it would impact other projector manufacturers from developing their own LED projectors (i.e how easy would it be to circumvent NEC's claims since idea as a whole wasn't new when they filed)

Mr Wiggles (geez thats too weird to type in a serious thought...)

I think assembly twelve is a typical projector optical assembly. And based on the patents I've looked at for other companies (InFocus, RCA, NEC, etc) the part count is normal. Hopefully, optics wouldn't drive the price through the roof.

That said, a previous thread pointed out that LED-based projectors could use plastic lenses since LED light lacks the IR/UV components.

Oh and anyone else guessing a major marketing point would be; "No fragile bulb to break when you carry it between meetings!" The optics would become the most delicate part. NEC makes a lot of business projectors, don't they?

Happy patent hunting

Davandron,

You are correct the light for optics before the imager can be plastic for the reasons you mentioned, but those after the imager I think still will need to be glass.

It will definitely be simpler than current LCD projectors but at least a little more complicated than current 1-chip DLP projectors.

The advantages are huge if they can get the light output up. (even if they can't it will still be nice for high gain screens etc.)

-Mr. Wigggles

The neat thing about getting the output up from an LED is that as lpw goes up, the junction temperature goes down. That means you can pack more devices into the same square area, which improves the amount of light you can get through the optical system. Osram and Lumileds have both demonstrated red LEDs with output of more 100 lpw. I'd be curious to know just how efficient NEC's optics are.

Even the best coated glass is only 99% efficient at the boundaries where air meets glass (there are some internal losses as well). A more conservative number is about 97% (meaning 3% of the light is reflected back at each boundary.)

A 10% efficient DLP projector is pretty good when it is all said and done with conventional color wheels etc. Lumileds said that they should be able to get 36% effecient DLP in the PDF above.

The key element is green LEDs. The Red LEDs are plenty bright at this point. It is the Green that are going to need some further devlopement. The way lumens work is that you need about 3X the Green lumens as you need Red and about 3X the Red lumens as you need blue.

So hopefully a 5 Watt Red LED, 20 Watts of Green LEDs, and about 20 Watts of Blues would be all we need for 1000 lumen projector. Put in another 50 Watts for all the processing we could want and we are at about 100 Watts total.

Hopefully.

-Mr. Wigggles

Agfa has been using the exact same technology on there Hi speed photographic printer for 6 years. it is used to print index prints, and it does work very good and is reliable. here is the link. Dale

http://www.agfa.com/photo/products/l..._print_systems

Wiggles,

When you talk about lumens of red/green/blue, you're referring to the human/eye response, right?

As I understood it, sensitivity went from Red at the lowest, to Blue at the highest. But perhaps you are adjusting for the very high efficiency red LEDs, and talking about power consumption not output when you talk about watts?

Here is some info on the human eye and color theory for everyone's advantage.

What are NEC's likely business plans?

It seems that most new DLP technology ends up in RPTVs first. Will NEC follow this path, or will they move into front projection? Light output may already be in the ballpark for RPTVs, and if NEC combined LEDs with TI's xHD3 chip I think they could have a very high performance product.

Any guesses as to what this could sell for?

Dave

We as humans are senstive to green, red and then blue in that order.

When I talk about Watts it is in reference to the Watts going to the LEDs not the radiometric output of the LEDs.

Lumens are not radiometric (i.e. not Watts) they are based on the humans ability to detect output. Humans are much more senstive to green than they are red and even more to green vs. blue.

lumens are based on this physiology. Blue in ANY projector is almost always putting out more actual Watts than red or green.

-Mr. Wigggles

Wigggles, even the bright DLPs like the Infocus 7200 are about 8% efficient (200w UHP @ 60lpw = 12,000 lumens). Something like a Sharp 12K appears to be about 6%. And that is starting with a really small arc. It's hard to tell from NEC's patent if they've solved the etendue problem, but I think LED projectors will need a lot more light that you're supposing. I don't foresee LEDs outshining discharge projectors for as long time to come at equivalent power levels. The main advantage will be the elimination of the color wheel, so the RGB sequence can be sped up, and fractionally more mirror time for less dithering. In addition, the sources will likely last the life of the projector. The real difficulty will be keeping the LEDs cool, as life and output decline drastically with junction temperature. Packaged LEDs like the Luxeons are not the answer, as they have a relatively high thermal resistivity.

scoby,

One of the key benefits of LED is that you can run them in burst mode for a single chip design.

When I say a 5 Watt red LED, what I really mean is a 5 Watt red LED run at 15 Watts 1/3rd of the time.

This is where Lumileds 36% efficiency claim comes from. You don't scrub 67+% of your light with a colorwheel like you do with a LAMP based design.

I said "hopefully" because I don't know how accurate Lumiled's numbers are simply that I understand where they are coming from.

-Mr. Wigggles

EDIT: Ohlson, yes I meant lamp.

MrW
You mean ___ like you do with a lamp based design. ?

If LEDs have a lower lumens per watt and are bigger than the arc of a traditionla lamp why are they so promising? We want delivery in two years and not in twenty years.

For everyone's enjoyment, here is a technical summary of some of LEDs advantages, a few of which we've been talking about:

1. Direct control of the light source
1a. "No Color Wheel:" The color wheel currently absorbs a significant amount of light to enable the system to be controlled. It presents an inefficiency that LEDs will not have to overcome

1b. "No more rainbows:" Since the colors can be cycled extremely quickly, much quicker than mechanical systems, the rainbow problem should no longer be observable by anyone.

Additional information on this topic can be found below as Appendix A

1c "No fixed color order, duration, or intensity" Currently, the DMD / DLP system works by having a constant know light source and modulating a mirror to adjust the intensity of each pixel. The video processor must work with a fixed color sequence, with each color having a fixed duration of type to create the image. Put directly, the system must make use of 100% Red for 2.5mS then 100% Green for 2.5mS then 100% Blue for 2.5mS.

An LED system does not contain this limit. The video processor can control the light based on the properties of the image it is attempting to display. It can control the color, the brightness, and the duration of any color the LED array can produce. Again, put directly, the processor could order up 90% Red for 1mS then 100% Blue for 2.0mS then 20% Green for 0.5mS then 50% Green AND 50% Red (a dim yellow) for 1.0mS then 100% Red and 100% Green and 100% Blue (pure white) for 1.8mS.... I hope everyone can see what I am getting at.

2. LEDs simplify cooling.
2a. "No IR / UV component of light" Current lamps produce more than just visible light, they produce infrared and ultraviolet. These must be filtered and also create a significant amount of heat which propagates through the optical system. An LED system lacks this concern and should create less cooling demand for the optics (quieter fans).

2b. "Thermal energy can be cooled directly." LEDs can have heatsinks attached to them directly to aid in the cooling of the devices without interfering with their light output. This should enable more efficient cooling (quieter fans).

3. Service life between 10,000 and 100,000 hours. With an extended service life, you would never have to replace a bulb. A MTBF (Mean time before failure) of 10,000 hours is considered the benchmark for consumer electronics; for a tv thats watching 4 hours a day, everyday for over 6 years.

4. LEDs are more robust than existing bulbs. LEDs can withstand far greater mechanical shocks than bulbs can and should mean a sturdier mobile projector.

5. Arc bulbs require a complex power control system to prevent the color of the light from changing. The power control system for LEDs is simplier.

6. Instant on LEDs do not need to warm-up and stabilize as arc-bulbs do and projectors using LEDs would turn on as fast as direct view LCDs.

Enjoy!

Appendix A: Continued 1b information
Note, the following is a fairly technical list of articles/quotes and I apologize that it is not intriniscially obvious why this proves LEDs can elliminate rainbows.

From this article, the following quote is used:

This equates to a subframe lasting for 400uS. Also please note that a subframe is not a complete image, but the longest time that one color is displayed before switching to a new color.

From this DLP White paper

That gives a switching window of 24uS. This is significantly faster than our mimimum subframe duration.

Although I won't site any sources (I leave it up to the critic to disprove) LEDs can switch in less than 0.1uS, but even at 10uS they still more than fast enough.

[Edit: Added 1b Info]

Wigggles, while a lot of T1-3/4 LEDs have peak forward currents much higher than their rated forward current, this isn't true of the Luxeon parts. I don't have any data sheets in front of me, but I think the 350mA parts have a peak forward current of 500mA. Besides, you don't really want to do this. One advantage of having each LED running only 1/3rd of the time is that keeps the junction temperature to a more manageable level. The problem is not creating the light, though. The problem is collecting and collimating it, and getting it through an optical system. Larger sources mean larger optics or less efficient systems. I'm not sure that this particular nut has been cracked yet.